Using AI and ARC to think like the other side
When geopolitical analysts wrestle with a complex geopolitical question, say
the future of European financial support to Ukraine, they tend to begin with
familiar sources: Western media, think tank reports, and official statements.
But what if, instead, they ran the same scenario-building exercise with Russian
sources threaded in? The same Russian sources a Russian analyst might leverage
as part of their own analysis? What might change in the narratives, framing,
and assessed likelihoods and what could we learn by studying where there was
divergence or consistency?
The concept of putting yourself in the place of your adversary is not new, but I was curious to experiment with ARC to see what AI would generate and if this was a capability we could build upon in the future. While I ran into a couple small hiccups along the way, in minutes, I had generated a table of scenarios and likelihood ranges generated from two different source pools (in multiple languages), with highlights of the differences between each comparable scenario. Surely a professional geopolitical analyst could nitpick these and would need to massage them, but in just a few minutes I was able to have a "draft" contrasting how “our” side and “their” side might tell the story of the same potential future to provide a lens into how adversaries perceive our vulnerabilities and opportunities.
Running the Experiment with ARC
I began with a research question: "What triggers could result in European aid to Ukraine transitioning from grants to loans in the next 3 years?"
ARC began collecting open sources, a vast majority of which were from U.S. and Western Europe news outlets and think tanks. When it had completed its initial search, I asked it to first do a decomposition of drivers and signals, then produce 4 possible outcome scenarios (not mutually exclusive or exhaustive). Critically for this experiment, when I create scenarios I can choose to leverage not just the LLM (in this case ChatGPT) but also the high-relevance sources ARC has found for me from the web. I checked this box, and in a few seconds had 4 scenarios to begin working with.
Next, I replicated this research question and replaced the sources ARC found with sources I manually found posing as a Russian speaker sitting in Russia using Yandex (Russia's Google equivalent) news search. I tried to confirm the sources I had found were from independent reporting inside Russia, and ran the same decomposition and scenario development exercises, this time asking ARC to only use the saved Russian sources I had found.
The results were telling. Russian-sourced scenarios consistently emphasized European weakness, fiscal limits, and public fatigue, portraying loan pivots as reluctant necessities. Western-sourced scenarios, by contrast, emphasized institutional mechanisms, policy frameworks, and reform benchmarks, portraying shifts as deliberate and managed.
Even the likelihood ranges diverged. Russian scenarios tended to give higher probabilities to outcomes in which Europe and the U.S. falter, while Western scenarios reserved higher likelihoods for adaptive continuity.
Why Shift Your Perspective?
Experts in intelligence analysis have been encouraging studying your adversary's point of view for a long time. Sherman Kent, in Strategic Intelligence for American World Policy (1949), stressed the need to model adversary decision-making rather than mirror one’s own assumptions. Robert Jervis’s Perception and Misperception in International Politics (1976) highlighted how states systematically misread others by failing to adopt their perspective.
Today this is formalized as "red teaming" or adversary modeling—techniques designed to anticipate strategy by “thinking like the enemy” (Zenko, Red Team, 2015). There are also Structured Analytic Techniques (SATs), such as Devil’s Advocacy and Alternative Competing Hypotheses (Heuer & Pherson, Structured Analytic Techniques for Intelligence Analysis, 2010) to give analysts ways to deliberately role-play an adversary’s worldview.
What AI adds is the speed to potentially do this across whatever perspective you want to take. Instead of a human red team laboring over foreign press and policy statements, AI can quickly structure scenarios, extract implicit likelihoods, and highlight divergences in framing with any permutation of sources it has access to.
Implications for Policymakers
Adversaries are not neutral observers. They are shaping their own narratives for their own audiences and for us. By comparing analysis derived from sources an adversary is leveraging with our own, policymakers can:
-
See how they are being seen: Understanding what vulnerabilities
adversaries highlight (e.g., “European fatigue,” “budgetary squeeze”) helps
stress-test our own policies. Does their outside perspective actually have
merit we should consider?
-
Anticipate adversary strategy: If Russian narratives consistently
stress Western weakness for example, this may signal where Moscow intends to
try and exploit.
-
Craft resilient communication: Policymakers can inoculate the
public against adversary narratives by preemptively addressing them.
- Design better policy options: Recognizing where adversaries expect cracks can help strengthen coalitions, reinforce institutions, and close off lines of manipulation and disinformation.
Doing analytic exercises to think like your adversary has been doctrine for decades. But with AI, we can completely change the timescale, consistency, and scope of performing this exercise. By running scenario exercises for distinct information ecosystems for example, analysts and policymakers can quickly map not just what they believe about the future, but what their adversaries believe about the future and shape their policies and decisions accordingly.
---
My scenarios comparison
These scenarios were generated by ARC, with no intervention on my part to adjust language or likelihoods. The assessed difference was generated by ChatGPT.
What triggers could result in European aid to Ukraine transitioning from grants to loans in the next 3 years?
| Scenarios (Western sources) | Scenarios (Russian sources) |
|---|---|
|
Gradual
Fiscal Shift: Loans Become the Dominant Form of EU Aid
Likelihood: 55% to 80% Persistent fiscal pressures within major EU economies, combined with increased defense and social spending, lead to a steady reduction in EU grant-based aid to Ukraine. As grant fatigue grows among both policymakers and the public, and as the EU seeks to maintain support for Ukraine without expanding its budget, the majority of new macro-financial support transitions to concessional loans. The shift is formalized through new EU legislative and financial frameworks, such as the expanded Ukraine Loan Cooperation Mechanism, and is justified as prudent fiscal management. Grants are reserved for humanitarian needs and select reform-linked projects, while mainstream budget support and reconstruction funding are issued as loans—often backed by future revenue from immobilized Russian assets. This transition is accompanied by increased conditionality and debt sustainability monitoring by the IMF and EU institutions. |
Budgetary
Squeeze: European Fiscal Pressures Drive Shift from Grants to Loans
Likelihood: 20% - 45% A combination of recession, competing domestic crises (migration, energy, disaster relief), and growing political fatigue within Europe erodes willingness and ability to sustain large-scale grant-based aid. Parliamentary debates and EU budget negotiations highlight the limits of current spending, with public calls to prioritize domestic needs. To maintain support for Ukraine while addressing fiscal constraints, the EU restructures future aid as predominantly loans, with only targeted grants for critical sectors. |
| Assessed Difference: The Russian-sourced scenario frames the shift as a reluctant consequence of European weakness, crisis, and political fatigue. The Western-sourced scenario frames it as a deliberate, institutionalized policy adjustment for fiscal prudence, with sustained commitment to Ukraine. The Russian version is shorter, vaguer, and more negative; the Western version is detailed, technical, and emphasizes continuity of support through structured mechanisms. | |
|
Conditionality Backlash: Governance and Reform Failures Trigger Loan-Only Support Likelihood: 20% - 45% A significant deterioration in Ukraine’s governance, with missed EU/IMF reform benchmarks and high-profile corruption scandals, leads to a loss of confidence among EU member states. Amid growing donor fatigue and pressure from skeptical publics, the EU formally suspends most grant-based aid, citing unmet anticorruption and judicial reform requirements. Instead, any continued financial support is restructured as loans with strict conditionality and delayed disbursement schedules. This hardline shift is justified as necessary to protect EU taxpayer funds and to incentivize Ukrainian reforms. Humanitarian assistance continues as grants, but budgetary and reconstruction support are exclusively loan-based until reform progress is verified. |
Conditionality Crunch: Aid Tied to Governance Reforms Triggers Shift to Loans Likelihood: 45% - 55% Ukraine's continued failure to meet EU anti-corruption and governance reform benchmarks leads to a hardening of EU conditionality. After multiple warnings, partial suspensions, and reductions of grant tranches, the EU transitions most new financial support to loans, arguing that only demonstrated reform progress will justify future grants. The transition is publicly justified as a means to incentivize reforms and protect European taxpayer interests, with some humanitarian and military aid remaining as grants. |
|
Assessed Difference: The Russian-sourced scenario emphasizes process, gradual tightening, and fair discipline, with the shift to loans framed as pragmatic and taxpayer-oriented. The Western-sourced scenario emphasizes scandal, crisis, and backlash, with a dramatic loss of confidence leading to an aid cutoff. The Russian version is more procedural and balanced, while the Western version is harsher, more urgent, and portrays the shift as a political rupture rather than a managed adjustment. |
|
|
Geopolitical Shock: US Withdrawal and Security Escalation Force EU Loan Pivot Likelihood: 45% - 55% A major escalation in the Ukraine conflict (e.g., a renewed Russian offensive or significant territorial losses) coincides with a formal and lasting US withdrawal of non-military aid to Ukraine. Facing the collapse of US budgetary support, the EU is compelled to fill the gap to prevent Ukrainian state failure. However, the scale of the required financial commitment, combined with rising EU defense costs and domestic opposition, makes continued grant-based aid politically unsustainable. The EU rapidly transitions most macro-financial support to loan format, using new mechanisms (e.g., Russian asset-backed loans), while maintaining minimal grants for humanitarian needs. This transition is abrupt and primarily driven by exogenous geopolitical events. |
US Disengagement: Loss of American Support Forces Europe to Restructure Aid Terms Likelihood: 20% - 45% A US administration (or Congress) dramatically reduces or ends financial support to Ukraine and publicly insists Europeans take over the burden. Facing much larger funding gaps and unable to sustain grants at scale, the EU transitions the bulk of its assistance to loans, arguing that it is the only way to maintain support at required volumes. The shift is presented as necessary fiscal realism given the absence of transatlantic burden-sharing. |
|
Assessed Difference: The Russian-sourced scenario emphasizes Europe’s fiscal weakness and dependence on US burden-sharing, framing the loan shift as pragmatic realism after Washington exits. The Western-sourced scenario emphasizes geopolitical crisis and urgency, framing the loan pivot as an abrupt, externally forced necessity to prevent collapse. The Russian version is budget-centric and downplays external events; the Western version is crisis-centric, with security escalation as the key trigger. |
|
|
Sustained Grants: EU Maintains Current Aid Structure Despite Pressures Likelihood: 20% - 45% Despite fiscal, political, and geopolitical pressures, EU member states maintain a balanced mix of grants and loans to Ukraine through 2028. Strong public and elite support for Ukraine persists in key EU countries, and Ukraine makes visible progress on reforms and anticorruption benchmarks. The EU is able to mobilize new own resources (e.g., via new taxes or expanded joint borrowing) and unlocks more revenue from Russian assets, allowing continued grant-based support. US policy stabilizes at a low but nonzero level, reducing the risk of an abrupt EU funding pivot. The Ukraine Facility continues to disburse grants for macro-financial and reconstruction support alongside concessional loans, with no major change in the overall aid mix. |
Status Quo Sustained: Grants Remain Primary Mode Despite Internal and External Pressures Likelihood: 55% - 80% Despite ongoing governance, fiscal, and political challenges, the EU and its member states continue to provide most Ukraine aid as grants. Incremental reforms by Ukraine, European unity in the face of Russian aggression, and successful management of internal budgetary pressures allow the current model to persist. Loans remain a supplementary tool, but grants dominate due to political symbolism, humanitarian needs, and the desire to avoid saddling Ukraine with long-term debt. |
|
Assessed Difference: The Russian-sourced scenario frames sustained grant support as probable and natural, justified by unity, humanitarian imperatives, and symbolism. The Western-sourced scenario frames it as less likely, requiring resource innovation, reform progress, and public support to maintain, with grants continuing but only alongside structured loan mechanisms. The Russian framing is simpler and more political, while the Western framing is more technical, detailed, and conditional. |
|
